Add initial Set implementation to tests, see #17; Fix check when shrinking a Map

std/assembly/arraybuffer.ts

@@ -32,11 +32,11 @@ export class ArrayBuffer {
 
   static readonly HEADER_SIZE: usize = HEADER_SIZE;
 
-  @inline load<T>(index: i32): T {
+  @inline load<T>(index: usize): T {
     return load<T>(changetype<usize>(this) + index * sizeof<T>(), HEADER_SIZE);
   }
 
-  @inline store<T>(index: i32, value: T): void {
+  @inline store<T>(index: usize, value: T): void {
     store<T>(changetype<usize>(this) + index * sizeof<T>(), value, HEADER_SIZE);
   }
 }

tests/compiler/std/map.ts

@@ -2,7 +2,7 @@ import {
   hash
 } from "internal/hash";
 
-// A deterministic hash table based on CloseTable from https://github.com/jorendorff/dht
+// A deterministic hash map based on CloseTable from https://github.com/jorendorff/dht
 
 const INITIAL_CAPACITY = 4;
 const FILL_FACTOR: f64 = 8 / 3;
@@ -19,6 +19,9 @@ class MapEntry<K,V> {
 /** Empty bit. */
 const EMPTY: usize = 1 << 0;
 
+/** Size of a bucket. */
+const BUCKET_SIZE = sizeof<usize>();
+
 /** Computes the alignment of an entry. */
 @inline function ENTRY_ALIGN<K,V>(): usize {
   // can align to 4 instead of 8 if 32-bit and K/V is <= 32-bits
@@ -51,7 +54,7 @@ class Map<K,V> {
   constructor() { this.clear(); }
 
   clear(): void {
-    const bucketsSize = INITIAL_CAPACITY * <i32>sizeof<usize>();
+    const bucketsSize = INITIAL_CAPACITY * <i32>BUCKET_SIZE;
     this.buckets = new ArrayBuffer(bucketsSize);
     this.bucketsMask = INITIAL_CAPACITY - 1;
     const entriesSize = INITIAL_CAPACITY * <i32>ENTRY_SIZE<K,V>();
@@ -71,29 +74,27 @@ class Map<K,V> {
   }
 
   has(key: K): bool {
-    return this.find(key, hash(key)) !== null;
+    return this.find(key, hash<K>(key)) !== null;
   }
 
   get(key: K): V {
-    var entry = this.find(key, hash(key));
+    var entry = this.find(key, hash<K>(key));
     return entry ? entry.value : <V>unreachable();
   }
 
   set(key: K, value: V): void {
-    var hashCode = hash(key);
+    var hashCode = hash<K>(key);
     var entry = this.find(key, hashCode);
     if (entry) {
       entry.value = value;
     } else {
       // check if rehashing is necessary
-      let capacity = this.entriesCapacity;
-      if (this.entriesOffset == capacity) {
+      if (this.entriesOffset == this.entriesCapacity) {
         this.rehash(
-          this.entriesCount >= <i32>(capacity * FREE_FACTOR)
-            ? (this.bucketsMask << 1) | 1 // grow capacity to next 2^N
-            :  this.bucketsMask           // just rehash if 1/4+ entries are empty
+          this.entriesCount < <i32>(this.entriesCapacity * FREE_FACTOR)
+            ?  this.bucketsMask           // just rehash if 1/4+ entries are empty
+            : (this.bucketsMask << 1) | 1 // grow capacity to next 2^N
         );
-        capacity = this.entriesCapacity;
       }
       // append new entry
       let entries = this.entries;
@@ -102,48 +103,51 @@ class Map<K,V> {
       );
       entry.key = key;
       entry.value = value;
+      ++this.entriesCount;
       // link with previous entry in bucket
       let bucketIndex = hashCode & this.bucketsMask;
       entry.taggedNext = this.buckets.load<usize>(bucketIndex);
       this.buckets.store<usize>(bucketIndex, changetype<usize>(entry));
-      ++this.entriesCount;
     }
   }
 
   delete(key: K): bool {
-    var entry = this.find(key, hash(key));
+    var entry = this.find(key, hash<K>(key));
     if (!entry) return false;
     entry.taggedNext |= EMPTY;
     --this.entriesCount;
+    // check if rehashing is appropriate
+    var halfBucketsMask = this.bucketsMask >> 1;
     if (
-      this.bucketsMask  > <u32>INITIAL_CAPACITY &&
-      this.entriesCount < <i32>(this.entriesOffset * FREE_FACTOR)
-    ) this.rehash(this.bucketsMask >> 1);
+      halfBucketsMask + 1 >= max<u32>(INITIAL_CAPACITY, this.entriesCount) &&
+      this.entriesCount < <i32>(this.entriesCapacity * FREE_FACTOR)
+    ) this.rehash(halfBucketsMask);
     return true;
   }
 
-  private rehash(newBucketsMask: i32): void {
-    var newBucketsCapacity = newBucketsMask + 1;
-    var newBuckets = new ArrayBuffer(newBucketsCapacity * sizeof<usize>());
+  private rehash(newBucketsMask: u32): void {
+    var newBucketsCapacity = <i32>(newBucketsMask + 1);
+    var newBuckets = new ArrayBuffer(newBucketsCapacity * <i32>BUCKET_SIZE);
     var newEntriesCapacity = <i32>(newBucketsCapacity * FILL_FACTOR);
-    var newEntries = new ArrayBuffer(newEntriesCapacity * ENTRY_SIZE<K,V>(), true);
+    var newEntries = new ArrayBuffer(newEntriesCapacity * <i32>ENTRY_SIZE<K,V>(), true);
 
     // copy old entries to new entries
-    var p = changetype<usize>(this.entries) + ArrayBuffer.HEADER_SIZE;
-    var q = changetype<usize>(newEntries) + ArrayBuffer.HEADER_SIZE;
-    var k = p + this.entriesOffset * ENTRY_SIZE<K,V>();
-    while (p != k) {
-      let pEntry = changetype<MapEntry<K,V>>(p);
-      let qEntry = changetype<MapEntry<K,V>>(q);
-      if (!(pEntry.taggedNext & EMPTY)) {
-        qEntry.key = pEntry.key;
-        qEntry.value = pEntry.value;
-        let bucketIndex = hash(pEntry.key) & newBucketsMask;
-        qEntry.taggedNext = newBuckets.load<usize>(bucketIndex);
-        newBuckets.store<MapEntry<K,V>>(bucketIndex, qEntry);
-        q += ENTRY_SIZE<K,V>();
+    var oldPtr = changetype<usize>(this.entries) + ArrayBuffer.HEADER_SIZE;
+    var oldEnd = oldPtr + <usize>this.entriesOffset * ENTRY_SIZE<K,V>();
+    var newPtr = changetype<usize>(newEntries) + ArrayBuffer.HEADER_SIZE;
+    while (oldPtr != oldEnd) {
+      let oldEntry = changetype<MapEntry<K,V>>(oldPtr);
+      if (!(oldEntry.taggedNext & EMPTY)) {
+        let newEntry = changetype<MapEntry<K,V>>(newPtr);
+        newEntry.key = oldEntry.key;
+        newEntry.value = oldEntry.value;
+        let newBucketIndex = hash<K>(oldEntry.key) & newBucketsMask;
+        let newBucketPtr = changetype<usize>(newBuckets) + <usize>newBucketIndex * BUCKET_SIZE;
+        newEntry.taggedNext = load<usize>(newBucketPtr, ArrayBuffer.HEADER_SIZE);
+        store<usize>(newBucketPtr, newPtr, ArrayBuffer.HEADER_SIZE);
+        newPtr += ENTRY_SIZE<K,V>();
       }
-      p += ENTRY_SIZE<K,V>();
+      oldPtr += ENTRY_SIZE<K,V>();
     }
 
     this.buckets = newBuckets;
@@ -160,54 +164,55 @@ function test<K,V>(): void {
   var map = new Map<K,V>();
 
   // insert new
-  for (let k: K = 1; k <= 200; ++k) {
-    map.set(k, 100 + <V>k);
+  for (let k: K = 0; k < 100; ++k) {
+    assert(!map.has(k));
+    map.set(k, 10 + <V>k);
     assert(map.has(k));
-    assert(!map.has(k + 1));
-    assert(map.get(k) == 100 + k);
+    assert(map.get(k) == 10 + <V>k);
   }
-  assert(map.size == 200);
+  assert(map.size == 100);
 
   // insert duplicate
-  for (let k: K = 50; k <= 100; ++k) {
+  for (let k: K = 0; k < 100; ++k) {
     assert(map.has(k));
-    assert(map.get(k) == 100 + <V>k);
-    map.set(k, 100 + <V>k);
+    assert(map.get(k) == 10 + <V>k);
+    map.set(k, 20 + <V>k);
     assert(map.has(k));
-    assert(map.get(k) == 100 + <V>k);
+    assert(map.get(k) == 20 + <V>k);
   }
-  assert(map.size == 200);
+  assert(map.size == 100);
 
   // delete
-  for (let k: K = 1; k <= 100; ++k) {
+  for (let k: K = 0; k < 50; ++k) {
     assert(map.has(k));
-    assert(map.get(k) == 100 + <V>k);
+    assert(map.get(k) == 20 + <V>k);
     map.delete(k);
     assert(!map.has(k));
-    assert(map.has(k + 1));
   }
-  assert(map.size == 100);
+  assert(map.size == 50);
 
   // insert + delete
-  for (let k: K = 1; k <= 50; ++k) {
+  for (let k: K = 0; k < 50; ++k) {
     assert(!map.has(k));
-    map.set(k, 100 + <V>k);
+    map.set(k, 10 + <V>k);
     assert(map.has(k));
     map.delete(k);
     assert(!map.has(k));
   }
-  assert(map.size == 100);
+  assert(map.size == 50);
 
   // clear
   map.clear();
   assert(map.size == 0);
 }
 
-test<i32,i32>();
-test<i64,i32>();
-test<i64,i64>();
-test<i32,i64>();
+test<i8,i32>();
+test<u8,i32>();
 test<i16,i32>();
-test<i16,i64>();
-test<i32,i16>();
-test<i64,i16>();
+test<u16,i32>();
+test<i32,i32>();
+test<u32,i32>();
+test<i64,i32>();
+test<u64,i32>();
+test<f32,i32>();
+test<f64,i32>();

tests/compiler/std/set.ts

@@ -1,30 +1,204 @@
+import {
+  hash
+} from "internal/hash";
+
+// A deterministic hash set based on CloseTable from https://github.com/jorendorff/dht
+
+const INITIAL_CAPACITY = 4;
+const FILL_FACTOR: f64 = 8 / 3;
+const FREE_FACTOR: f64 = 3 / 4;
+
+/** Structure of a set entry. */
+@unmanaged
+class SetEntry<K> {
+  key: K;
+  taggedNext: usize; // LSB=1 indicates EMPTY
+}
+
+/** Empty bit. */
+const EMPTY: usize = 1 << 0;
+
+/** Size of a bucket. */
+const BUCKET_SIZE = sizeof<usize>();
+
+/** Computes the alignment of an entry. */
+@inline function ENTRY_ALIGN<K>(): usize {
+  // can align to 4 instead of 8 if 32-bit and K is <= 32-bits
+  const align = (sizeof<K>() > sizeof<usize>() ? sizeof<K>() : sizeof<usize>()) - 1;
+  return align;
+}
+
+/** Computes the aligned size of an entry. */
+@inline function ENTRY_SIZE<K>(): usize {
+  const align = ENTRY_ALIGN<K>();
+  const size = (offsetof<SetEntry<K>>() + align) & ~align;
+  return size;
+}
+
+class Set<K> {
+
+  // buckets holding references to the respective first entry within
+  private buckets: ArrayBuffer; // usize[bucketsMask + 1]
+  private bucketsMask: u32;
+
+  // entries in insertion order
+  private entries: ArrayBuffer; // SetEntry<K>[entriesCapacity]
+  private entriesCapacity: i32;
+  private entriesOffset: i32;
+  private entriesCount: i32;
+
+  get size(): i32 { return this.entriesCount; }
+
+  constructor() { this.clear(); }
+
+  clear(): void {
+    const bucketsSize = INITIAL_CAPACITY * <i32>BUCKET_SIZE;
+    this.buckets = new ArrayBuffer(bucketsSize);
+    this.bucketsMask = INITIAL_CAPACITY - 1;
+    const entriesSize = INITIAL_CAPACITY * <i32>ENTRY_SIZE<K>();
+    this.entries = new ArrayBuffer(entriesSize, true);
+    this.entriesCapacity = INITIAL_CAPACITY;
+    this.entriesOffset = 0;
+    this.entriesCount = 0;
+  }
+
+  private find(key: K, hashCode: u32): SetEntry<K> | null {
+    var entry = this.buckets.load<SetEntry<K>>(hashCode & this.bucketsMask);
+    var i = 0;
+    while (entry) {
+      if (!(entry.taggedNext & EMPTY) && entry.key == key) return entry;
+      entry = changetype<SetEntry<K>>(entry.taggedNext & ~EMPTY);
+    }
+    return null;
+  }
+
+  has(key: K): bool {
+    return this.find(key, hash(key)) !== null;
+  }
+
+  add(key: K): void {
+    var hashCode = hash(key);
+    var entry = this.find(key, hashCode);
+    if (!entry) {
+      // check if rehashing is necessary
+      if (this.entriesOffset == this.entriesCapacity) {
+        this.rehash(
+          this.entriesCount < <i32>(this.entriesCapacity * FREE_FACTOR)
+            ?  this.bucketsMask           // just rehash if 1/4+ entries are empty
+            : (this.bucketsMask << 1) | 1 // grow capacity to next 2^N
+        );
+      }
+      // append new entry
+      let entries = this.entries;
+      entry = changetype<SetEntry<K>>(
+        changetype<usize>(entries) + ArrayBuffer.HEADER_SIZE + this.entriesOffset++ * ENTRY_SIZE<K>()
+      );
+      entry.key = key;
+      ++this.entriesCount;
+      // link with previous entry in bucket
+      let bucketIndex = hashCode & this.bucketsMask;
+      entry.taggedNext = this.buckets.load<usize>(bucketIndex);
+      this.buckets.store<usize>(bucketIndex, changetype<usize>(entry));
+    }
+  }
+
+  delete(key: K): bool {
+    var entry = this.find(key, hash<K>(key));
+    if (!entry) return false;
+    entry.taggedNext |= EMPTY;
+    --this.entriesCount;
+    // check if rehashing is appropriate
+    var halfBucketsMask = this.bucketsMask >> 1;
+    if (
+      halfBucketsMask + 1 >= max<u32>(INITIAL_CAPACITY, this.entriesCount) &&
+      this.entriesCount < <i32>(this.entriesCapacity * FREE_FACTOR)
+    ) this.rehash(halfBucketsMask);
+    return true;
+  }
+
+  private rehash(newBucketsMask: u32): void {
+    var newBucketsCapacity = <i32>(newBucketsMask + 1);
+    var newBuckets = new ArrayBuffer(newBucketsCapacity * <i32>BUCKET_SIZE);
+    var newEntriesCapacity = <i32>(newBucketsCapacity * FILL_FACTOR);
+    var newEntries = new ArrayBuffer(newEntriesCapacity * <i32>ENTRY_SIZE<K>(), true);
+
+    // copy old entries to new entries
+    var oldPtr = changetype<usize>(this.entries) + ArrayBuffer.HEADER_SIZE;
+    var oldEnd = oldPtr + <usize>this.entriesOffset * ENTRY_SIZE<K>();
+    var newPtr = changetype<usize>(newEntries) + ArrayBuffer.HEADER_SIZE;
+    while (oldPtr != oldEnd) {
+      let oldEntry = changetype<SetEntry<K>>(oldPtr);
+      if (!(oldEntry.taggedNext & EMPTY)) {
+        let newEntry = changetype<SetEntry<K>>(newPtr);
+        newEntry.key = oldEntry.key;
+        let newBucketIndex = hash<K>(oldEntry.key) & newBucketsMask;
+        let newBucketPtr = changetype<usize>(newBuckets) + <usize>newBucketIndex * BUCKET_SIZE;
+        newEntry.taggedNext = load<usize>(newBucketPtr, ArrayBuffer.HEADER_SIZE);
+        store<usize>(newBucketPtr, newPtr, ArrayBuffer.HEADER_SIZE);
+        newPtr += ENTRY_SIZE<K>();
+      }
+      oldPtr += ENTRY_SIZE<K>();
+    }
+
+    this.buckets = newBuckets;
+    this.bucketsMask = newBucketsMask;
+    this.entries = newEntries;
+    this.entriesCapacity = newEntriesCapacity;
+    this.entriesOffset = this.entriesCount;
+  }
+}
+
 import "allocator/arena";
 
-// note that this doesn't test a real set implementation yet, see std/assembly/set.ts
+function test<K>(): void {
+  var set = new Set<K>();
 
-var set = changetype<Set<i32>>(allocate_memory(sizeof<usize>() + 2 * sizeof<i32>()));
+  // insert new
+  for (let k: K = 0; k < 100; ++k) {
+    assert(!set.has(k));
+    set.add(k);
+    assert(set.has(k));
+  }
+  assert(set.size == 100);
 
-assert(set.size == 0);
+  // insert duplicate
+  for (let k: K = 50; k < 100; ++k) {
+    assert(set.has(k));
+    set.add(k);
+    assert(set.has(k));
+  }
+  assert(set.size == 100);
 
-set.add(1);
-set.add(0);
-set.add(2);
+  // delete
+  for (let k: K = 0; k < 50; ++k) {
+    assert(set.has(k));
+    set.delete(k);
+    assert(!set.has(k));
+  }
+  assert(set.size == 50);
 
-assert(set.size == 3);
+  // insert + delete
+  for (let k: K = 0; k < 50; ++k) {
+    assert(!set.has(k));
+    set.add(k);
+    assert(set.has(k));
+    set.delete(k);
+    assert(!set.has(k));
+  }
+  assert(set.size == 50);
 
-assert(set.has(1));
-assert(set.has(0));
-assert(set.has(2));
-assert(!set.has(3));
+  // clear
+  set.clear();
+  assert(set.size == 0);
+}
 
-set.delete(0);
-
-assert(set.size == 2);
-assert(set.has(1));
-assert(!set.has(0));
-assert(set.has(2));
-
-set.clear();
-
-assert(set.size == 0);
-assert(!set.has(1));
+test<i8>();
+test<u8>();
+test<i16>();
+test<u16>();
+test<i32>();
+test<u32>();
+test<i64>();
+test<u64>();
+test<f32>();
+test<f64>();